Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. A system for a brain-computer interface, comprising: a remote device, and an implantable device comprising: an integrated circuit comprising: a signal filtering unit comprising a preamplifier and a filter, and configured to amplify and filter one or more extracellular recordings, an analogue-digital conversion unit configured to convert the one or more extracellular recordings, and a spike detection unit configured to detect, within the one or more extracellular recordings, at least one action potential generated by one or more neurons; and a template matching module arranged to: compare the at least one detected action potential with one or more predetermined spike templates, and only in response to a comparison that does not indicate a positive match between the at least one detected action potential and the one or more predetermined spike templates, transmit, to the remote device that is remote from the implantable device, information regarding the at least one detected action potential; wherein the remote device comprises: a template building module configured to generate (i) the one or more predetermined spike templates, and (ii) one or more new spike templates based on the information regarding the at least one detected action potential; wherein the implantable device is configured to receive the one or more new spike templates from the remote device.
A brain-computer interface system includes an implantable device and a remote device. The implantable device records extracellular neural signals, amplifies and filters them using a preamplifier and filter, converts them from analog to digital, and detects action potentials (spikes) generated by neurons. A template matching module compares detected spikes against stored spike templates. If no match is found, the implantable device transmits information about the unmatched spike to the remote device. The remote device includes a template building module that generates new spike templates based on the transmitted information and sends these back to the implantable device. This system dynamically updates spike templates to improve neural signal decoding accuracy over time. The implantable device operates autonomously, only communicating with the remote device when new spike templates are needed, reducing power consumption and data transmission. The remote device handles template generation, allowing the implantable device to focus on signal processing and detection. This approach enhances the adaptability of brain-computer interfaces to changing neural activity patterns.
2. The system according to claim 1 , wherein the template building module is arranged to wirelessly communicate with the implantable device.
This invention relates to a system for wireless communication with an implantable medical device, addressing the challenge of securely and efficiently transmitting data between an external template building module and an implanted device. The system includes a template building module that generates and manages communication templates for the implantable device, ensuring reliable data exchange. The module is designed to wirelessly interface with the implantable device, enabling real-time adjustments and monitoring of the device's operation. The communication templates may include protocols for data transmission, encryption standards, and power management settings to optimize performance and battery life. The system may also incorporate error detection and correction mechanisms to maintain data integrity during wireless communication. Additionally, the template building module can dynamically update the implantable device's firmware or configuration parameters based on received data, enhancing adaptability. The wireless communication may utilize low-power protocols to minimize energy consumption while ensuring robust connectivity. This approach improves the efficiency and reliability of interactions between external systems and implantable medical devices, supporting applications such as remote patient monitoring and therapeutic adjustments.
3. The system according to claim 1 , wherein the signal filtering unit filters the one or more extracellular recordings using one or more of an elliptic filter, a Butterworth filter, a Bessel filter, or a 2nd order filter.
This invention relates to a signal processing system for analyzing extracellular recordings, such as those obtained from neural or biological signals. The system addresses the challenge of accurately filtering these recordings to remove noise and artifacts while preserving the integrity of the desired signal components. Extracellular recordings often contain unwanted interference, which can distort the analysis of neural activity or other biological processes. Effective filtering is essential for extracting meaningful information from these signals. The system includes a signal filtering unit that processes the extracellular recordings using one or more specialized filters. These filters are selected from a set of high-performance options, including elliptic filters, Butterworth filters, Bessel filters, or second-order filters. Each filter type has distinct characteristics: elliptic filters provide sharp roll-off with minimal phase distortion, Butterworth filters offer a maximally flat frequency response, Bessel filters ensure minimal phase distortion, and second-order filters provide a balance between simplicity and performance. The filtering unit applies one or more of these filters to the recordings, allowing for customizable noise reduction tailored to the specific requirements of the application. This approach ensures that the filtered signals retain their fidelity while effectively suppressing unwanted noise, enabling more accurate analysis of the underlying biological or neural activity.
4. The system according to claim 1 , wherein a high-pass filter cut-off frequency of the signal filtering unit is about 300 Hz such that a performance value of the spike detection unit is maximized.
This invention relates to a signal processing system for detecting neural spikes in biological signals, such as those recorded from neural tissue. The system addresses the challenge of accurately identifying neural spikes while minimizing noise interference, which can degrade detection performance. The system includes a signal filtering unit that applies a high-pass filter to the input signal to remove low-frequency noise and baseline drift. The filter's cut-off frequency is set to approximately 300 Hz to optimize the performance of a subsequent spike detection unit. The spike detection unit processes the filtered signal to identify neural spikes based on predefined criteria, such as amplitude thresholds or waveform shapes. The system may also include a noise reduction unit to further enhance signal quality before spike detection. By tuning the high-pass filter to 300 Hz, the system maximizes the detection accuracy and reliability of the spike detection unit, ensuring that relevant neural activity is captured while irrelevant low-frequency components are suppressed. This approach improves the overall efficiency and accuracy of neural signal analysis in applications such as brain-machine interfaces, neuroprosthetics, and neural research.
5. The system according to claim 1 , wherein a low-pass filter cut-off frequency of the signal filtering unit is about 3 kHz such that a performance value of the spike detection unit is maximised.
This invention relates to a signal processing system for optimizing spike detection in neural signal analysis. The system addresses the challenge of accurately identifying neural spikes in recorded bioelectric signals, which are often corrupted by noise and high-frequency artifacts. The core innovation involves a signal filtering unit that applies a low-pass filter with a cut-off frequency of approximately 3 kHz to preprocess the input signal before spike detection. This specific cut-off frequency is selected to maximize the performance of a subsequent spike detection unit, ensuring that relevant neural activity is preserved while minimizing interference from irrelevant high-frequency noise. The filtering unit processes the raw input signal, removing unwanted frequency components, and the optimized cut-off frequency enhances the signal-to-noise ratio for downstream spike detection. The spike detection unit then analyzes the filtered signal to identify neural spikes with improved accuracy and reliability. The system is particularly useful in applications such as brain-machine interfaces, neural prosthetics, and neurophysiological research, where precise spike detection is critical for interpreting neural activity.
6. The system according to claim 1 , wherein a sampling rate of the analogue-digital conversion unit is reduced to about 7 kHz such that a performance of the spike detection unit remains above 90%.
The invention relates to a system for neural signal processing, specifically for optimizing the performance of a spike detection unit in a neural recording system. The system includes an analogue-digital conversion unit that converts neural signals from an analogue format to a digital format. A key challenge in such systems is balancing power consumption and processing efficiency while maintaining high accuracy in detecting neural spikes. The invention addresses this by reducing the sampling rate of the analogue-digital conversion unit to approximately 7 kHz. This reduction is carefully calibrated to ensure that the performance of the spike detection unit remains above 90%, meaning that the system retains high accuracy in identifying neural spikes despite the lower sampling rate. The spike detection unit processes the digitized signals to identify and analyze neural activity patterns. By optimizing the sampling rate, the system achieves energy efficiency without compromising detection performance, making it suitable for low-power, long-term neural monitoring applications. The invention is particularly useful in implantable or wearable devices where power consumption is a critical constraint.
7. The system according to claim 1 , wherein a resolution of the analogue-digital conversion unit is 6 bits or 10 bits such that a performance of the spike detection unit remains above 90%.
The system relates to a neural signal processing system designed to detect and analyze neural spikes with high accuracy. The system includes an analogue-digital conversion unit that converts neural signals from analog to digital form. The resolution of this conversion unit is specifically set to either 6 bits or 10 bits to ensure that the performance of the subsequent spike detection unit remains above 90%. The spike detection unit processes the digitized signals to identify neural spikes, which are brief electrical events generated by neurons. The system is optimized to maintain high detection accuracy despite variations in signal resolution, ensuring reliable neural signal analysis. This configuration is particularly useful in applications such as brain-machine interfaces, neuroprosthetics, and neural research, where precise spike detection is critical for interpreting neural activity. The system balances resolution and performance to achieve efficient and accurate neural signal processing.
8. The system according to claim 1 , wherein a resolution of the analogue-digital conversion unit is 6 bits or 10 bits such that a performance of the template building module remains above 50%.
The invention relates to a system for processing signals, particularly focusing on improving the performance of a template building module within an analogue-digital conversion system. The system addresses the challenge of maintaining high accuracy in template generation despite variations in the resolution of the analogue-digital conversion unit. The analogue-digital conversion unit converts analogue signals into digital signals with a resolution of either 6 bits or 10 bits. The template building module processes these digital signals to generate templates used for further signal analysis or recognition tasks. The system ensures that the performance of the template building module remains above 50% when operating with either 6-bit or 10-bit resolution. This is achieved by optimizing the template building process to handle the lower resolution (6 bits) without significant degradation in performance, while also accommodating the higher resolution (10 bits) for more precise applications. The system is designed to be flexible, allowing it to adapt to different resolution settings while maintaining reliable template generation. This ensures consistent performance across varying signal processing requirements.
9. The system according to claim 1 , wherein the spike detection unit is configured to automatically select an amplitude threshold based on an estimate of a standard deviation of a background noise in the one or more extracellular recordings.
This invention relates to a system for detecting neural spikes in extracellular recordings, addressing the challenge of accurately identifying spike events while minimizing false positives caused by background noise. The system includes a spike detection unit that automatically adjusts an amplitude threshold for spike detection based on an estimate of the standard deviation of the background noise in the recordings. This adaptive thresholding approach improves detection accuracy by dynamically accounting for variations in noise levels, ensuring reliable spike identification even in noisy environments. The system may also include preprocessing components to filter and amplify the extracellular signals before spike detection, enhancing signal quality. Additionally, the spike detection unit may apply additional criteria, such as timing or waveform shape, to further refine spike identification. The overall system is designed for use in neural signal processing applications, such as brain-machine interfaces or neurophysiological research, where precise spike detection is critical for interpreting neural activity.
10. The system according to claim 9 , wherein the standard deviation of the background noise is estimated based on a median of an absolute value of the filtered one or more extracellular recordings.
The invention relates to a system for processing extracellular recordings, such as neural signals, to estimate background noise. In neural signal processing, accurately estimating background noise is critical for improving signal quality and extracting meaningful biological information. Existing methods may struggle with noise estimation in dynamic or non-stationary environments, leading to inaccuracies in signal analysis. The system processes one or more extracellular recordings, which are filtered to remove unwanted artifacts or frequencies. The filtered recordings are then used to estimate the standard deviation of background noise. Specifically, the noise estimation is based on the median of the absolute values of the filtered recordings. This approach leverages the statistical properties of the data to provide a robust noise estimate, reducing sensitivity to outliers or transient disturbances. The system may also include additional components, such as a filter for preprocessing the recordings and a processor for computing the noise estimate. The filtered recordings are analyzed to compute the absolute values, and the median of these values is calculated to derive the noise standard deviation. This method improves noise estimation accuracy, particularly in noisy or variable recording conditions, enhancing the reliability of subsequent signal analysis. The system is applicable in neuroscience, biomedical engineering, and other fields requiring precise noise characterization in extracellular recordings.
11. The system according to claim 1 , wherein the template building module comprises: a feature extraction unit configured to calculate a set of features for each of the at least one detected action potential; and a spike classification unit configured to generate the one or more predetermined spike templates from one or more principal features shared by a cluster of action potentials having similar features.
This invention relates to a system for processing neural signals, specifically for classifying action potentials (spikes) detected from neural recordings. The system addresses the challenge of accurately identifying and categorizing neural spikes, which is essential for applications such as brain-machine interfaces, neuroprosthetics, and neural signal analysis. The system includes a template building module that enhances spike classification by extracting and analyzing key features from detected action potentials. The template building module comprises a feature extraction unit that calculates a set of features for each detected action potential. These features may include waveform characteristics, timing information, or other relevant metrics that distinguish different types of spikes. The module also includes a spike classification unit that generates one or more predetermined spike templates based on principal features shared by clusters of action potentials with similar characteristics. By grouping spikes with comparable features, the system can create representative templates that improve the accuracy and efficiency of spike classification. This approach allows the system to adapt to variations in neural signals, ensuring reliable identification of different spike types even in noisy or dynamic recording environments. The use of principal features and clustering techniques enables the system to handle diverse neural activity patterns, making it suitable for a wide range of neurophysiological applications.
12. The system according to claim 11 , wherein the feature extraction unit is configured to at least one of: extract one or more data points for each detected action potential, calculate a plurality of principal components, or calculate a wavelet transform comprising a plurality of wavelet coefficients and select a preferred set from the plurality of wavelet coefficients based on a calculated deviation from normality according to a Kolmogorov-Smirnoff test.
This invention relates to a system for analyzing neural signals, specifically action potentials, to extract meaningful features for further processing. The system addresses the challenge of accurately identifying and characterizing neural activity patterns from raw signal data, which is often noisy and complex. The feature extraction unit processes detected action potentials by performing one or more of the following operations: extracting key data points from each action potential, computing principal components to reduce dimensionality and highlight significant variations, or applying a wavelet transform to decompose the signal into frequency components. In the wavelet transform approach, the system calculates multiple wavelet coefficients and selects a subset based on statistical deviation from normality, using a Kolmogorov-Smirnoff test to identify the most relevant coefficients. This selection process enhances the system's ability to distinguish between different neural activity patterns by focusing on the most informative signal components. The extracted features can then be used for tasks such as neural signal classification, decoding, or monitoring. The system improves upon prior methods by providing flexible, statistically robust feature extraction techniques tailored to neural signal analysis.
13. The system according to claim 12 , wherein the spike classification unit is configured to at least one of: detect the cluster of action potentials by applying a superparamagnetic clustering algorithm based on at least one of: the preferred set of wavelet coefficients, the plurality of principal components, or the extracted one or more data points, or detect the cluster of action potentials used to generate the one or more predetermined spike templates based on a cluster size criterion.
This invention relates to a system for classifying neural signals, specifically action potentials, in neural data processing. The system addresses the challenge of accurately detecting and classifying neural spikes from raw electrophysiological recordings, which is critical for applications in brain-machine interfaces, neuroprosthetics, and neural research. The system includes a spike classification unit that processes neural signals to identify clusters of action potentials, which are then used to generate spike templates for classification. The spike classification unit employs advanced signal processing techniques to analyze neural data. It can detect clusters of action potentials using a superparamagnetic clustering algorithm, which leverages either wavelet coefficients, principal components, or extracted data points from the neural signals. Alternatively, the unit may detect clusters based on a cluster size criterion, ensuring that only significant groupings of action potentials are considered. This approach enhances the accuracy and reliability of spike classification by refining the detection process before template generation. The system improves upon prior methods by incorporating flexible clustering techniques and size-based filtering, reducing false positives and improving the fidelity of neural signal interpretation. This enables more precise neural decoding and better performance in applications requiring real-time neural data analysis.
14. The system according to claim 1 , wherein the template matching module compares the at least one detected action potential with the one or more predetermined spike templates using one of a squared Euclidean distance metric, a Norm 1 distance metric, a Norm infinite distance metric, a Mahalanobis distance metric, and a nearest neighbours distance metric.
This invention relates to a system for analyzing neural signals, specifically for detecting and classifying action potentials (spikes) in recorded neural data. The system addresses the challenge of accurately identifying and categorizing neural spikes by comparing detected action potentials against predefined spike templates using various distance metrics. The template matching module evaluates the similarity between detected spikes and stored templates using one of several mathematical distance measures: squared Euclidean distance, Norm 1 distance, Norm infinite distance, Mahalanobis distance, or nearest neighbors distance. These metrics quantify the dissimilarity between the detected spike waveform and the template waveforms, enabling precise classification. The system enhances the reliability of neural signal analysis by providing multiple distance measurement options, allowing for flexibility in matching accuracy depending on the application requirements. This approach improves the robustness of spike sorting, which is critical for applications in neuroscience research, brain-machine interfaces, and medical diagnostics. The use of different distance metrics ensures adaptability to varying signal characteristics and noise levels, leading to more accurate neural data interpretation.
15. The system according to claim 1 , wherein the template matching module comprises a peak alignment unit configured to substantially align the at least one detected action potential with the one or more predetermined spike templates, to within a predetermined peak-alignment error.
The system relates to neural signal processing, specifically for analyzing action potentials (spikes) detected from neural recordings. The problem addressed is accurately matching detected action potentials to predefined spike templates, which is challenging due to variations in signal shape, timing, and noise. The system includes a template matching module with a peak alignment unit that aligns detected action potentials with one or more predetermined spike templates. The alignment is performed to within a predetermined peak-alignment error, ensuring precise matching despite signal variability. This alignment process helps distinguish between different neural sources by comparing the shape and timing of the detected spikes to the templates. The system may also include other components, such as a signal preprocessing unit to filter and amplify neural signals, and a spike detection unit to identify action potentials from the raw data. The peak alignment unit enhances the accuracy of spike sorting by compensating for small timing differences between the detected spikes and the templates, improving the reliability of neural signal analysis. This technology is useful in neuroscience research, brain-machine interfaces, and medical applications where precise neural signal decoding is required.
16. The system according to claim 1 , wherein the template matching module is configured to use a window size of about 0.5 ms.
The invention relates to a system for template matching in signal processing, particularly for analyzing time-domain signals such as those in radar, communications, or biomedical applications. The system addresses the challenge of accurately detecting and matching signal patterns within noisy or dynamic environments by employing a template matching module with a precisely defined window size. The template matching module compares a reference signal template against segments of an input signal to identify matches, and the specified window size of approximately 0.5 milliseconds ensures high-resolution alignment between the template and the input signal. This configuration enhances detection accuracy by minimizing timing errors and improving the system's ability to resolve closely spaced signal features. The system may also include preprocessing modules to condition the input signal, such as filtering or normalization, and post-processing modules to refine match results, such as thresholding or statistical analysis. The overall system is designed to operate in real-time or near-real-time applications where rapid and precise signal pattern recognition is critical. The invention improves upon prior art by optimizing the window size for specific signal characteristics, reducing false positives, and increasing detection reliability.
17. The system according to claim 1 , further comprising a selector configured to enable a user to manually select an amplitude threshold for the spike detection unit.
This invention relates to a signal processing system designed to detect and analyze electrical spikes, such as those generated by neural activity. The system addresses the challenge of accurately identifying and measuring spikes in noisy or variable signal environments, which is critical for applications in neuroscience, medical diagnostics, and bioelectronic devices. The core functionality involves a spike detection unit that processes input signals to identify spikes based on predefined amplitude thresholds. To enhance flexibility and user control, the system includes a selector that allows users to manually adjust the amplitude threshold used by the spike detection unit. This feature enables fine-tuning of detection sensitivity, accommodating different signal characteristics or experimental conditions. The selector may be implemented as a physical or software-based interface, providing real-time adjustments without requiring system recalibration. By integrating this manual threshold control, the system improves adaptability and accuracy in spike detection, ensuring reliable performance across diverse applications. The invention builds on foundational spike detection techniques by adding user-configurable parameters, making it more versatile for research and clinical use.
Unknown
November 3, 2020
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.